Nonideal Statistical Rate Theory Formulation To Predict Evaporation

Oct 27, 2008 - Cite this:J. Phys. Chem. B 112, 47, 15005-15013 ... Aaron H. Persad and Charles A. Ward. Chemical ... Atam Kapoor and Janet A. W. Ellio...
0 downloads 0 Views 135KB Size
J. Phys. Chem. B 2008, 112, 15005–15013

15005

Nonideal Statistical Rate Theory Formulation To Predict Evaporation Rates from Equations of State Atam Kapoor and Janet A. W. Elliott* Department of Chemical and Materials Engineering, UniVersity of Alberta, Edmonton, Alberta, Canada T6G 2G6 ReceiVed: June 5, 2008; ReVised Manuscript ReceiVed: July 28, 2008

A method of including nonideal effects in the statistical rate theory (SRT) formulation is presented and a generic equation-of-state based SRT model was developed for predicting evaporation rates. Further, taking the Peng-Robinson equation of state as an example, vapor phase pressures at which particular evaporation rates are expected were calculated, and the predictions were found to be in excellent agreement with the experimental observations for water and octane. A high temperature range (near the critical region) where the previously existing ideal SRT model is expected to yield inaccurate results was identified and predictions (for ethane and butane) were instead made with the Peng-Robinson based SRT model to correct for fluid nonidealities at high temperatures and pressures. 1. Introduction Recent experimental measurements have found that the temperature on the vapor side of an evaporating liquid is higher than the temperature on the liquid side. Temperature jumps of 5-8 K have been observed.1 This can not be explained by models based on classical kinetic theory2 which predict liquid evaporation only if the vapor temperature is less than the temperature on the liquid side. This is because all of these models assume the boundary condition of the temperature of the molecules evaporating to be the temperature of the liquid and thus the solutions generated are opposite to the experimental observations. Since classical kinetic theory models were shown to fail, quantum mechanical models of isolated thermodynamic systems were developed to explain the observations. A new model called statistical rate theory (SRT),3-5 based on the concept of transition probability, was used to investigate the observations. This model uses the Boltzmann definition of entropy and the equilibrium exchange rate to obtain the SRT expression for predicting kinetic rates of various processes in terms of measurable molecular and material variables. SRT assumes the transport processes to be primarily single-molecular events and is then developed using a first-order perturbation analysis of the Schro¨dinger equation. To date, over 75 papers have been published on this theory. SRT has been used in the past to predict rates for various processes ranging from gas absorption at liquid-gas interfaces,6 gas adsorption on solid substrates,4,5,7-10 electron transfer between ions in solution,4,11 migration of CO across a stepped Pt(111) surface,12 crystal growth from solution,13 and transport across cell membranes.14,15 Ward and Fang16,17 used the SRT model to predict the evaporation rates of water and hydrocarbons and explained the unexpected temperature jump across the interface. However, in all these applications, the SRT model has always been implemented by assuming ideality of gaseous phases and the incompressibility of solid/liquid phases. Even the experiments measuring the temperature jump1 across the interface have * Corresponding author. Phone: 1 780 492 7963. Fax: 1 780 492 2331. E-mail: [email protected].

Figure 1. Schematic of a steady-state evaporation system (adapted from Figure 1 in ref 16). We assume that the macroscopic curvature only affects the pressure difference across the interface; however, in a subvolume, Gibbs dividing surface approximation has been applied.

been at sufficiently low temperatures and pressures that the ideal SRT model was in close agreement with the experimental results. This is because the nonideal and compressibility effects could be accurately neglected and so no attempt was made to incorporate fluid bulk-phase nonidealities in the SRT expressions. In this paper, a method to incorporate nonidealities using equations of state in the SRT model is presented and a general expression for any equation of state is given. Further, the Peng-Robinson equation of state (fairly accurate for hydrocarbons) has been taken as an example and predictions have been made using both the Peng-Robinson SRT model and the ideal SRT model to identify the region where the two models differ. The rest of the paper is organized as follows: In section 2, we present a review of the SRT approach for predicting evaporation rates. In section 3, the ideal assumptions made by Ward and Fang16,17 to predict evaporation rates are presented. In section 4, we introduce the equation of state in the SRT model and develop a generic evaporation rate expression. In section 5, we present the results and make further predictions to identify the range where nonideal effects are expected. Finally, in section 6 concluding remarks are presented. 2. Statistical Rate Theory for Evaporation: A Review In section 2, we review the development of SRT3,4 and specifically SRT applied to evaporation.16,17

10.1021/jp804982g CCC: $40.75  2008 American Chemical Society Published on Web 10/28/2008

15006 J. Phys. Chem. B, Vol. 112, No. 47, 2008

Kapoor and Elliott

2.1. Evaporation Rate Expression. Consider an isolated C-component thermodynamic system containing a liquid bulk phase and a vapor bulk phase with a liquid-vapor interface in between. Using the Gibbs dividing surface approximation for the interface, it is assumed that a molecule can only be in the liquid or the vapor phase. Suppose that the system is not in equilibrium i.e. the number of molecules in the vapor and the liquid are different from their equilibrium values. Consider an isolated element of the system obtained by dividing the system into small but finite subvolumes as shown in Figure 1. Let R and β denote the liquid and the vapor phases in the subvolume, respectively. Let, at any instant t, the molecular configuration of the system, λj be given as

λj(NR, Nβ):

(N1R, N2R, N3R, ..., NγR, ..., NCR), (Nβ1 , Nβ2 , Nβ3 , ..., Nβγ, ..., NβC)

where NRi and Nβi are the number of molecules of component i in the liquid and the vapor phase, respectively. SRT predicts the rate at which component γ is transferred from the liquid phase to the vapor phase. ˆ 0 and H ˆ denote the Hamiltonian operator for the system Let H when molecules are in the state λj, i.e., confined in their subvolume phases (unperturbed state), and when the molecules are free to move across the interface between the subvolume phases (perturbed state), respectively. We can define a potential ˆ -H ˆ 0, which allows the molecules to interact operator Vˆ ≡ H with their actual potential. The Schro¨dinger equation for the system when the molecules are allowed to move freely across the interface becomes

p ∂ψ ˆ 0 + Vˆ)ψ ) (H i ∂t

(1)

where ψ is the wave function for the liquid-vapor system. To ˆ 0ψ| is very apply perturbation theory, we assume that |Vˆψ|/|H small; however, this assumption is different from saying that Vˆ is zero, because in that case the molecules would stay in their confined state, thereby making the transition probability zero. Considering the energy uncertainty principle, eigenfunctions xr0 corresponding to the unperturbed Hamiltonian within an energy uncertainty ∆E can be determined from

ˆ 0xr0 ) Er0xr0 H

tion similar to Gibbs “principle of equal a priori probabilities” in statistical thermodynamics, however, here applied to a thermodynamic system not in equilibrium. At this instant, the system has a transition probability to move to a molecular configuration as a result of the transfer of one molecule (of component γ) from the liquid phase R to the vapor phase β. We also assume at this point that the number of molecules in other subvolumes does not change during this transition. This configuration λk will be given as

λk(NR - 1, Nβ + 1): (N 1R, N 2R, N 3R, ..., N γR - 1, ..., N CR), (N β1 , N β2 , N β3 , ..., N βγ + 1, ..., N βC) If the number of quantum mechanical states corresponding to this molecular distribution is given by Ω(λk), then according to the standard perturbation analysis, the probability of a transition from molecular distribution λj to λk is given as

τ(λj f λk) ) K(λj, λk)

Ω(λk) Ω(λj)

(4)

where

K(λj, λk) )

2π ω |V | 2 p j mn

(5)

where ωj and Vmn are the microscopic state density of molecular configuration λj and matrix elements of the operator Vˆ, respectively. At the same instant, there is a possibility of a single molecular transition in the opposite sense, to the molecular configuration, λp.

λp(NR + 1, Nβ - 1): (N 1R, N 2R, N 3R, ..., N γR + 1, ..., N CR), (N β1 , N β2 , N β3 , ..., N βγ - 1, ..., N βC) The transition probability from λj to the configuration λp is given as

τ(λj f λp) ) K(λj, λp)

Ω(λp) Ω(λj)

(6)

(2) where

where E0 is the energy eigenvalue. These energy eigenvalues are highly degenerate. Corresponding to a molecular distribution λj, there are Ω(λj) quantum mechanical states. Thus, there are Ω(λj) wave functions which correspond to the same molecular distribution and so the wave function of the unperturbed liquid-vapor system can be written as a linear combination of all of the possible Ω(λj) wave functions as

K(λj, λp) )

2π ω |V | 2 p j ms

(7)

where ωj and Vms are the microscopic state density of molecular configuration λj and matrix elements of the operator Vˆ, respectively. Combining eqs 4 and 6, the net rate of the forward process (evaporation) can be written as

Ω(λj)

ψ)

Er t ∑ Crxr0 exp( -i p 0)

(3)

r

where Cr are linear coefficients. The system may be in any of the Ω(λj) quantum mechanical states. In writing this, we assume that the system is equally likely to be in any of the quantum mechanical states within an energy uncertainty, ∆E, an assump-

JRβ ) K(λj, λk)

Ω(λk) Ω(λp) - K(λj, λp) Ω(λj) Ω(λj)

(8)

Introducing the Boltzmann definition of entropy, S(λd) ) k ln Ω(λd), where k is the Boltzmann constant, the numbers of microstates in eq 8 can be replaced by the entropy of the states,

Evaporation Rates from Equations of State

J. Phys. Chem. B, Vol. 112, No. 47, 2008 15007

yielding the final expression for the rate of molecular transfer from phase R to β3,4

JRβ

[

]

S(λk) - S(λj) ) K(λj, λk) exp k

[

K(λj, λp) exp

]

S(λp) - S(λj) (9) k

2.2. Entropy Change Calculation. Since entropy is an extensive quantity, we can use the additive property of entropy to write the total entropy change during the transition as the sum of entropy changes for the liquid (R), vapor (β), and the reservoir (R)



S(λk) - S(λj) )

[Si(λk) - Si(λj)]

(10)

S(λp) - S(λj) ) -[S(λk) - S(λj)]

2.3. Equilibrium Molecular Exchange Rate. To develop thermodynamic expressions for the K’s, we consider the system in the limit of equilibrium when the entropy difference between the states before and after the transition approaches zero. Thus, the rate in eq 9 becomes

JRβ ) K(λe, λf) - K(λe, λb)

K(λe, λf) ) K(λe, λb) ) Ke

Using Euler relations,18 the entropy change in eq 10 may be written as16

(

(

HR(λk)

-

HR(λj)

)

+ TR TR Hβ(λj) HR(λj) Hβ(λk) HR(λk) µR µβ + R - β + Tβ Tβ T T TR TR (11)

) (

) (

)

where H, T, and µ are the enthalpy, temperature, and the chemical potential of the phase, respectively. It has been assumed that intensive thermodynamic variables remain unchanged as a result of a single molecular event. Since we are considering an isolated system, there is no net energy change during the molecular transition event. Thus, we can write the energy balance equation for the isolated system as

[HR(λk) - HR(λj)] + [Hβ(λk) - Hβ(λj)] + [HR(λk) - HR(λj)] ) 0 (12) Substituting eq 12 in eq 11 yields

S(λk) - S(λj) )

(

)

µR µβ + TR Tβ 1 1 - R [Hβ(λk) - Hβ(λj)] (13) β T T

(

)

The enthalpy per molecule h is now introduced by using Hi ) hiNi to replace [Hβ(λk) - Hβ(λj)] in eq 13 by hβ∆Nβ ) hβ (the transition causes a change of one molecule, so ∆Nβ ) 1) to obtain the final expression for entropy change as16

S(λk) - S(λj) )

(

(16)

where λe is the equilibrium molecular configuration, and λf and λb are molecular conifgurations on either side of the equilibrium configuration. At equilibrium, although the system is subject to these fluctuations in the molecular configuration, the net rate JRβ should be zero. Thus, from eq 16

i)R,β,R

S(λk) - S(λj) )

(15)

) (

µR 1 µβ 1 - β + hβ β - R R T T T T

)

(14)

Since the entropy change in the reverse process (from configuration λj to λp) is simply the negative of the entropy change in eq 14, we can write

(17)

This means that K(λe, λf) is independent of the direction of the transition and does not depend on the molecular configurations. Since, for molecular distributions near equilibrium, K(λj, λk) is independent of the distributions λj and λk, we can generalize this property of K to all near-equilibrium molecular transitions,3,4 i.e.,

K(λe+s, λe+s-1) ) K(λe+s-1, λe+s-2) ) .... ) K(λe+1, λe) ) K(λe, λe-1) ) K(λe-1, λe-2) ) .... ) K(λe-h, λe-h-1) ) Ke (18) where Ke is the equilibrium value. Since the K’s are independent of the initial and final states, they represent the reversible part in the rate expression in eq 9 and the entropy terms in the rate expression can be considered to give rise to irreversibility. Since at equilibrium, the unidirectional rate of molecular transport from phase R to phase β is Jf Rβ ) Ke and from phase f β to phase R is JβR ) Ke, it follows that Ke is the unidirectional equilibrium molecular exchange rate between the two phases. Thus, Ke is a function of the equilibrium properties of the system. Thus, substituting eqs 14, 15, and 18 into eq 9 gives the final expression for the rate for evaporation as16

{ [

JRβ ) Ke exp

(

)]

µR µβ hβ 1 1 + - R k Tβ kTR kTβ T µR µβ hβ 1 1 + - R exp k Tβ kTR kTβ T

[

(

)] }

(19)

3. Ideal SRT In section 3, we review the ideal treatment of the fluid phases in the previous application of SRT to evaporation.16 3.1. Ideal Gas Approximation. To evaluate the expression for net rate of evaporation obtained in the previous section (eq 19), we need expressions for the equilibrium molecular exchange rate (Ke), chemical potentials, and specific enthalpies.ConsideringBoltzmannstatisticsandtheBorn-Oppenheimer approximation, translational, vibrational, rotational, and electronic partition functions of an ideal gas can be written as19,20

15008 J. Phys. Chem. B, Vol. 112, No. 47, 2008

qtrans )

(

2πmkT h2

NV

qVib )

)

3/2

V)

V Λ3

Kapoor and Elliott

(20)

exp(-θ /2T)

l ∏ 1 - exp(-θ l/T)

(21)

l)1

qrot )

( ) 2kT h2

3/2 (πI)1/2

(22)

σs

( )

qelec ) geexp

De kT

(23)

where m is the molecular mass, V is the total volume of the ideal gas, and Λ is the de Broglie wavelength of the ideal gas particle; ge and De are the degeneracy of the energy state and the reference minimum potential for the electronic partition function qelec, respectively; θl and NV are the characteristic vibrational temperature and number of vibrational degrees of freedom of the particle, respectively; and I and σs are the moment of inertia and the symmetry number of the ideal gas molecule representing the number of indistinguishable orientations of the molecule, respectively. Since the total energy of the molecule can be written as the sum of translational, vibrational, rotational, and electronic energies, we can write the overall partition function, Q as the product of individual partition functions19,20

Q ) qtransqvibqrotqelec

(24)

Thus, the chemical potential and specific enthalpy can be evaluated as16,19

µid ∂ ln Q )kT ∂N

(

)

[( )

) -ln

V,T

m 2πh2

3/2 (kT)5/2

]

P ln(qvibqrotqelec) (25)

and

hid ∂ ln Q )T kT ∂T

(

)

V,N

+

De PV )4+ kT kT 3

Applying classical kinetic theory for the equilibrium state, we can determine the equilibrium exchange rate as the rate at which molecules from the vapor phase collide with the liquid-vapor interface assuming that all vapor molecules that collide with the interface successfully transfer to the liquid phase. Thus, at equilibrium conditions, the equilibrium exchange rate can be written as3,16

3

θ

∑ 2Tl + l)1

θ /T

l (26) ∑ exp(θl/T) -1

Pe

Ke )

(28)

√2πmkTe

where Pe and Te are the equilibrium pressure and temperature of the system. 3.3. Ideal SRT Expression. Substituting eqs 25-28 in eq 19, we can evaluate the evaporation rate in terms of known thermodynamic quantities as16,17

J)

Pβe

∆S -∆S exp( ) - exp( [ k k )] √2πmkT R

where

[( ) ] ( )

R qvib(Tβ) ∆S Tβ 4 P∞(T ) + ln + ) ln R k T Pβ qvib(TR) V∞ R (P - P∞(TR)) + kTR 3 θl θl 1 1 Tβ + 4 1 (30) + Tβ TR l)1 2 exp(θl/Tβ) - 1 TR

(

) ( )

)∑(

and where Pβe is the vapor phase pressure when the liquid-vapor system will be in equilibrium. Since at equilibrium the temperature of the isolated subvolume (and actually the entire system) will be the temperature of the liquid reservoir, Tβ ) TR, the temperature appearing as Te in eq 28 will be TR. Pβe can be evaluated by using the fact that chemical potentials of the liquid and vapor phase become equal at equilibrium. The chemical potential of the liquid phase is described by eq 27, and considering the vapor to be an ideal gas, its chemical potential can be written as

()

l)1

where V is the molecular volume, and the subscript “id” indicates that the expressions are valid only when ideal gas assumptions are appropriate. 3.2. Incompressible Liquid Treatment. The liquid phase is considered to be incompressible and an expression for its chemical potential can be found in terms of a reference state. If the saturation condition is chosen as its reference state, then

µR(TR, PR) ) µ[TR, P∞(TR)] + V∞R [PR - P∞(TR)] (27) where V∞R is the specific volume of the saturated liquid. The saturated state is also approximated as an ideal gas (just like the vapor phase) and its chemical potential is described by eq 25.

(29)

µβ(TR, Pβe ) ) µ[TR, P∞(TR)] + kTR ln

Pβe P∞

(31)

Equating eq 31 to eq 27, we get Pβe to be

(

Pβe ) P∞(TR) exp

V∞R kTR

)

[PRe - P∞(TR)]

(32)

In the experiments conducted and analyzed by Ward and Fang,1,16,17 evaporation flux was measured for different pressures in the vapor phase. The liquid temperature was fixed and the pressure in the liquid phase was determined by the Laplace equation

Evaporation Rates from Equations of State

PR ) Pβ +

2γ Rc

J. Phys. Chem. B, Vol. 112, No. 47, 2008 15009

(33)

where γ is the interfacial tension and Rc is the measured interface curvature. Thus, using eq 33 in eqs 30 and 32, the dependence of evaporation flux on liquid pressure is removed, and so eqs 29 and 30, which are now in terms of experimentally measured, or otherwise known, quantities only, can be used to predict vapor pressure for a particular flux to compare with the measured value of vapor pressure.16,17 Figure 2. Schematic diagram of a hypothetical surface.

4. Statistical Rate Theory: Nonideal Corrections Until now, SRT for evaporation has only been implemented using ideal approximations. Ward and Fang16,17 showed that eq 29 with eq 30 is in good agreement with the experimental results for water and octane when evaluated at pressures below 1 kPa and temperatures below room temperature. However, deviations from the experimental results are expected if we were to use this expression for high temperatures and high pressures where ideal gas approximations are no longer valid. The nonideality of gases at high temperatures and pressures can be quite significant. The objective of this paper is to develop an expression for the Statistical Rate Theory evaporation rate allowing for nonideality using equations of state. 4.1. Departure Functions for Chemical Potential and Enthalpy. For an equation of state

PV ) ZkT

(34)

with dimensionless compressibility Z ) Z(T, F), where F is the molar density, departure functions for chemical potential and enthalpy can be derived by taking the difference between the real fluid and the ideal gas property and integrating from the infinite volume condition (where the real fluid and the ideal gas are the same) to the actual volume of the system. Enthalpy and chemical potential departures can be written as21

µ - µid ) kT

∫0F (Z -F 1) dF + (Z - 1) - ln Z

h - hid ) kT

∂Z dF ∫0F -T( ∂T )F F + Z - 1

objective is to find the number of molecules crossing the surface S per unit time. Next, consider a differential element dx (big enough to assume a number density N/V), at a distance x from the surface S in the negative x direction, as shown in Figure 2. In one second, only those molecules will strike or cross the surface S, which have a velocity Vx g x. In a nonrarified gas, the molecules may collide with one another before striking the surface S. Since half of these moleculemolecule collisions will enhance the surface collisions, and half will reduce the surface collisions, without loss of generality we may neglect the impact of molecule-molecule collisions. Considering a Maxwell-Boltzmann distribution of velocities, we know that the probability of observing a molecule with a velocity Vx is given as22

(35)

p(Vx) )



(37)

and so we can say that the probability of Vx g x is equal to

p(Vx g x) )

∫x





[ ]

-mV2x m dVx exp 2πkT 2kT

(38)

Considering the number density and the volume of the differential element, we can write the total number of molecules in this differential element which may strike or cross the surface S, as

(36)

Adding eqs 35 and 36 appropriately to eqs 25 and 26, we can get the real enthalpy per molecule and chemical potential expressions based on the equation of state being obeyed in the range of temperature and pressure being considered. If the equation of state is valid for both vapor and liquid phases in this temperature and pressure range, then these chemical potential and enthalpy relations can be used for both the phases. 4.2. Nonideal Corrections to the Equilibrium Molecular Exchange Rate. The equilibrium exchange rate, Ke, is the rate at which molecules from the vapor phase collide with the liquid vapor interface when the system is in equilibrium. To evaluate Ke in terms of equilibrium thermodynamic variables, we need to develop an expression for the number of molecules which collide with the interface at given thermodynamic conditions. Consider a very large hypothetical box of volume V containing N molecules. Consider a hypothetical surface, S (area A) somewhere deep in the interior of the box. Our

[ ]

-mV2x m exp 2πkT 2kT

KdV e )

N (Adx)p(Vx g x) V

(39)

Substituting eq 38 into eq 39, integrating over all possible distances from the surface, A, we can find the total number of molecules striking or passing through the surface S per unit area per unit time as

Ke )

∫0 ∫x ∞



N V



[ ]

-mV2x m dVx dx exp 2πkT 2kT

(40)

Changing the order of integration (and appropriately adjusting the limits of integration), we can rewrite the above expression as

Ke )

∫0 ∫0 ∞

Vx

N V



[ ]

-mV2x m dx dυx exp 2πkT 2kT

Performing the integration, we will get

(41)

15010 J. Phys. Chem. B, Vol. 112, No. 47, 2008

Ke )

N V

Kapoor and Elliott

kT  2πm

(42)

which is a well-known result16,17 but we note it is equally valid for the nonideal case. Considering a general equation of state, we can replace N/V with P/ZkT from eq 34 to get the final expression for Ke in terms of pressure and temperature as

Ke )

1 Ze

Pe

(43)

√2πmkTe

Thus, we obtain the equilibrium molecular exchange rate in terms of equilibrium temperature, pressure and compressibility of the gas, the latter of which can be evaluated for a particular equation of state. Since eq 43 was derived for a Maxwell-Boltzmann distribution in the gas, Ze will be the compressibility of the gas phase (Zβ(Pβe , Tβe )) and as before Te ) TR. Therefore, we can evaluate expression 19 for high temperatures and pressures by using the nonideal corrections to chemical potential (eq 35), enthalpy (eq 36), and the equilibrium exchange rate (eq 43) in eq 19 to yield

Pβe 1 ∆S -∆S J) β exp - exp k k R Ze √2πmkT

(

)

Figure 3. Using the experimentally measured values of the radius of curvature and the temperatures on both sides of the interface for water (from Tables 1 and 2 in ref 16), vapor pressures have been predicted using the ideal16 and the Peng-Robinson based SRT model that would result in the experimentally measured evaporation flux. The predicted values of the vapor pressure for both models have been plotted on the ordinate, against the experimentally measured vapor pressure values on the abscissa. The points should lie on the 45°line, if they are in perfect agreement.

0.45724R2Tc2 a) Pc

(44)

b)

0.07780RTc Pc

R ) (1 + (0.37464 + 1.54226ω - 0.26992ω2)(1 - Tr0.5))2

[( ) ] ( )

R qvib(Tβ) ∆S Tβ 4 P∞(T ) + ln + ) ln R k T Pβ qvib(TR)

(

V∞

(

41-

1 1 (PR - P∞(TR)) + β - R R kT T T θl θl + + 2 exp(θl /Tβ) - 1

(

) [∫

R

)

)

where Vj is the molar volume, ω is the acentric factor and Tr is the reduced temperature, Tr ) T/Tc. In terms of compressibility defined by eq 34, eq 46 can be rearranged to give24

3



1)1

Z3 - (1 - B)Z2 + (A - 3B2 - 2B)Z (AB - B2 - B3) ) 0 (47)

]

T FR (Z - 1) + dFR + (ZR - 1) - ln ZR R 0 T FR β Fβ (Z - 1) dFβ + (Zβ - 1) - ln Zβ + 0 Fβ ∂Zβ dFβ Tβ Fβ -Tβ + Zβ - 1 1 - R (45) β β 0 ∂T Fβ F T β

[∫ [∫ ( )

](

]

)

4.3. An Example: The Peng-Robinson Equation of State. Since most hydrocarbons are well described using the PengRobinson equation of state, we take the Peng-Robinson equation of state as an example to predict evaporation rates for hydrocarbons. The Peng-Robinson equation of state is given by23

P)

RT aR - 2 Vj - b Vj + 2bVj - b2

(46)

where A ) aRP/R2T2 and B ) bP/RT. The integrals in eqs 35 and 36 can be evaluated for the Peng-Robinson equation of state to give:25

[

µ - µid ) Z - 1 - ln(Z - B) kT

[

(

A Z + (1 + √2)B ln B√8 Z + (1 - √2)B

(

)

h - hid ac(dR/dT) A × ) Z-1kT B√8 bR√8 Z + (1 + √2)B ln Z + (1 - √2)B

(

)]

(48)

)]

(49)

Thus, our final expression for the evaporation flux according to the Peng-Robinson equation of state is

Evaporation Rates from Equations of State

1 J) β Ze

J. Phys. Chem. B, Vol. 112, No. 47, 2008 15011

Pβe

∆S -∆S exp( ) - exp( ( k k )) √2πmkT

(50)

R

where

[( ) ] ( )

R qvib(Tβ) ∆S Tβ 4 P∞(T ) + ln + ) ln R k T Pβ qvib(TR) V∞ R 1 1 (P - P∞(TR)) + β - R R kT T T 3 θl θl + + 2 exp(θ /Tβ) - 1 l)1



(

(

(

) [

l

)

)

Tβ + ZR - 1 - ln(ZR - BR) TR AR ZR + (1 + √2)BR - Zβ - 1 - ln(Zβ - Bβ) ln BR√8 ZR + (1 - √2)BR Aβ Zβ + (1 + √2)Bβ + Zβ - 1 ln β β β B √8 Z + (1 - √2)B ac(dR/dT) Aβ Zβ + (1 + √2)Bβ Tβ ln 1 (51) TR Bβ√8 bR√8 Zβ + (1 - √2)Bβ 41-

(

(

(

)(

)] [

)] [

) ](

)

Pβe can be evaluated by using the fact that chemical potentials of the liquid and vapor phase become equal at equilibrium (µR(TR, PeR) ) µβ(TR, Peβ)). These chemical potentials can be obtained using the Peng-Robinson equation of state for both the phases to evaluate Pβe . However, this adds computational complexity to the iterative calculation of equilibrium vapor pressure that corresponds to a particular evaporation flux. Therefore, we again use eq 32 to evaluate Pβe , while we use the compressibility of the vapor phase and the saturation pressure (which appears in eq 32) evaluated using the Peng-Robinson equation of state. It must be noted that the need to evaluate Pβe as different from the saturation pressure arises only because of the curvature of the interface but eqs 50 and 51 are equally valid for flat interfaces. In the case of flat interfaces, Pβe is equal to the saturation pressure at the liquid temperature and can be evaluated using the Peng-Robinson equation of state, as we have done. Since the characteristic temperatures of vibration which appear in the vibrational partition function are not available for hydrocarbons, both the ideal SRT and Peng-Robinson based SRT expression are evaluated neglecting the vibrational terms in the final rate expressions. 5. Results 5.1. Comparison with Previous Experiments. As was indicated during the sensitivity analysis in refs 16 and 17, SRT calculations should be done to predict the vapor pressure for a particular evaporation flux rather than predicting evaporation rates for a particular vapor pressure. This is done because it was noted that evaporation flux can be predicted accurately only if the vapor pressure of hydrocarbons is measured to within 0.002 Pa (an unachievable accuracy). During the evaluation, the liquid-vapor interface was assumed to be spherical, resulting in a pressure difference (liquid pressure higher than the vapor pressure) of 2γ/Rc, where γ is the interfacial tension and Rc is the radius of curvature.

Figure 4. Using the experimentally measured values of the radius of curvature and the temperatures on both sides of the interface for octane (from Table 1 in ref 17), vapor pressures have been predicted using the ideal17 and the Peng-Robinson based SRT model that would result in the experimentally measured evaporation flux. The predicted values of the vapor pressure for both models have been plotted on the ordinate, against the experimentally measured vapor pressure values on the abscissa. The points should lie on the 45°line, if they are in perfect agreement.

Figure 5. Predictions for ethane at T ) 25 and 30 °C made using both the ideal and the Peng-Robinson based SRT model after assuming a zero temperature jump at the interface, have been shown. The predicted values of the vapor pressure for both the models have been plotted on the ordinate, against the evaporation flux on the abscissa. Deviations can be seen for high flux values.

As is evident from Figure 3, measurements and predictions based on both the ideal and the Peng-Robinson based SRT model are in excellent agreement with each other for evaporation of water at low pressures (below 1 kPa) and temperatures (below room temperature). At such experimental conditions, the nonideal additional terms in the Peng-Robinson based SRT model (eq 50 with eq 51) are negligible. Measurements and predictions for octane (shown in Figure 4) are also observed to be in close agreement and again predictions based on the ideal model are very close to the predictions based on the Peng-Robinson based SRT model. 5.2. Further Predictions Based on Peng-Robinson SRT. In order to identify the range where fluid nonidealities become significant and where the Peng-Robinson based SRT model would yield better results than the ideal SRT model, we make predictions for the evaporation flux at temperatures near the critical temperatures. Ethane (Tc ) 32.28 °C) and butane (Tc ) 152.05 °C) have been chosen because their critical temperature

15012 J. Phys. Chem. B, Vol. 112, No. 47, 2008

Figure 6. Predictions for butane at T ) 400 K made using both the ideal and the Peng-Robinson based SRT models after assuming a zero temperature jump at the interface, have been shown. The predicted values of the vapor pressure for both the models have been plotted on the ordinate, against the evaporation flux on the abscissa. Deviations can be seen for high flux values.

Kapoor and Elliott Similar results were also obtained for butane at 400 K and the vapor pressure predictions have been plotted against the evaporation flux in Figure 6. In all cases it was found that the Peng-Robinson based SRT model predicted a higher vapor pressure than the ideal SRT model for the same evaporation flux. As there is no way to a priori predict the temperature jump (it was a measured quantity in refs 1, 16, and 17), and as the temperature jump was found to decrease with the evaporation flux, we assumed no temperature jump in the above calculations. Further, a sensitivity analysis of the evaporation flux with the temperature jump across the interface was done by assuming a temperature difference of 2 K for butane, with the vapor side temperature higher than the liquid side temperature and the results are shown in Figure 7. It can be inferred that a minor temperature difference across the interface does not affect the predictions on vapor pressures, and so, plots of vapor pressures vs evaporation flux would look similar to the plots for zero temperature jumps. Thus, a minor temperature difference across the interface would make no significant difference in the predictions, or our conclusions. 6. Conclusions

Figure 7. The Peng-Robinson based SRT model was used to predict the vapor pressure for a liquid temperature of 400 K for butane assuming temperature jumps of 0 and 2 K across the interface. The predicted values of the vapor pressure for zero temperature difference across the interface are plotted on the ordinate, against the predicted vapor pressure values for a temperature jump of 2 K across the interface, on the abscissa. If they were in perfect agreement, they would lie on the 45° line.

values are low compared to other hydrocarbons and so experiments could be done easily. As was indicated in refs 1, 16, and 17, the temperature jump across the interface decreases with an increase in the vapor pressure, and we conclude that we can neglect the temperature jump across the interface while predicting evaporation flux at very high pressures and temperatures (in the near-critical region). Since the liquid vapor surface almost disappears, the surface tension is low and there is not much of a pressure difference; however we have still taken the pressure difference to be 2γ/Rc with a value of Rc ) 5 mm for all of the calculations. Surface tension values were taken from Hysis (v3.1, Hyprotech Ltd.; ethane: γ ) 0.12406 dyne/cm (at 30 °C), 0.51307 dyne/cm (at 25 °C), butane: γ ) 1.6548 dyne/cm (at 400 K)). Predictions were made at two different temperatures, T ) 25 and 30 °C for ethane with both the ideal and the Peng-Robinson based SRT models. The vapor pressure predictions from both the models have been plotted against evaporation flux in Figure 5. It can be observed that the vapor pressure decreases (almost linearly) with the evaporation flux, and the deviation between the predictions from both the models increases with evaporation flux.

A SRT formulation incorporating fluid nonidealities using equations of state was presented to develop a generic equation-of-state based SRT model for liquid evaporation. The Peng-Robinson equation-of-state based SRT model was found to predict evaporation rates as accurately as the ideal SRT model at low temperatures (below room temperature) and pressures (below 1 kPa) and was found to differ from the ideal model at sufficiently higher temperatures and pressures (near the critical temperature). The Peng-Robinson based SRT model predicted higher vapor pressures than the ideal SRT model irrespective of the temperature jump across the interface. Experiments should be done in the near-critical region (temperatures higher than 0.8Tc) to investigate the predictions that have been made. Acknowledgment. This research was funded by the Department of Chemical and Materials Engineering at the University of Alberta and the Natural Sciences and Engineering Research Council of Canada. J.A.W.E. holds a Canada Research Chair in Interfacial Thermodynamics. References and Notes (1) Fang, G.; Ward, C. A. Phys. ReV. E 1999, 59 (1), 417. (2) Schrage, R. W. In A Theoretical Study of Interphase Mass Transfer; Columbia University Press: New York, 1953; Chapter 3. (3) Ward, C. A.; Findlay, R. D.; Rizk, M. J. Chem. Phys. 1982, 76 (11), 5599. (4) Elliott, J. A. W.; Ward, C. A. In Equilibria and Dynamics of Gas Adsorption on Heterogenous Solid Surfaces; Elsevier: Amsterdam, 1997; Vol. 104, p 285. (5) Ward, C. A.; Elliott, J. A. W. Appl. Surf. Sci. 2002, 196, 202. (6) Ward, C. A.; Rizk, M.; Tucker, A. S. J. Chem. Phys. 1982, 76, 5606. (7) Ward, C. A.; Findlay, R. D. J. Chem. Phys. 1982, 76, 5615. (8) Findlay, R. D.; Ward, C. A. J. Chem. Phys. 1982, 76, 5624. (9) Elliott, J. A. W.; Ward, C. A. J. Chem. Phys. 1997, 106 (13), 5667. (10) Elliott, J. A. W.; Ward, C. A. J. Chem. Phys. 1997, 106 (13), 5677. (11) Ward, C. A. J. Chem. Phys. 1983, 79 (11), 5605. (12) Torri, M.; Elliott, J. A. W. J. Chem. Phys. 1999, 111 (4), 1686. (13) Dejmek, M.; Ward, C. A. J. Chem. Phys. 1998, 108, 8698. (14) Skinner, F. K.; Ward, C. A.; Bardakjian, B. L. BioPhys. J. 1993, 65, 618. (15) Elliott, J. A. W.; Elmoazzen, H. Y.; McGann, L. E. J. Chem. Phys. 2000, 113 (16), 6573. (16) Ward, C. A.; Fang, G. Phys. ReV. E 1999, 59 (1), 429. (17) Fang, G.; Ward, C. A. Phys. ReV. E 1999, 59 (1), 441.

Evaporation Rates from Equations of State (18) Callen, H. B. In Thermodynamics and an Introduction to Thermostatics, 2nd ed.; John Wiley and Sons, Inc.: New York, 1985; Chapter 3. (19) Hill, T. L. In An Introduction to Statistical Thermodynamics; Dover: New York, 1986; pp 74-80. (20) McQuarrie, D. A. In Statistical Mechanics; University Science Books: Sausalito, CA, 2003; pp 85-107. (21) Elliott, J. R.; Lira, C. T. In Introductory Chemical Engineering Thermodynamics; Prentice Hall: New York, 1999, p 235.

J. Phys. Chem. B, Vol. 112, No. 47, 2008 15013 (22) Bird, R. B.; Stewart, W. E.; Lightfoot, E. N. In Transport Phenomena; 2nd Ed., John Wiley and Sons, Inc.: New York, 2002, p 38. (23) Peng, D. Y.; Robinson, D. B. Ind. Eng. Chem.: Fundamentals 1976, 15, 58. (24) Elliott, J. R.; Lira, C. T. In Introductory Chemical Engineering Thermodynamics; Prentice Hall: New York,1999, p. 203. (25) Elliott, J. R.; Lira, C. T. In Introductory Chemical Engineering Thermodynamics; Prentice Hall: New York, 1999, p. 240-241.

JP804982G